Early intervention strategies in infant wheezing will be dependent on the ability to predict persistence of disease. We undertook a prospective longitudinal study to determine which factors might be predictive for the persistence of wheeze. We examined a group of 107 children 3 to 36 mo of age with at least one atopic parent. Children were recruited within 12 wk of first wheeze. Factors assessed included: personal atopy (IgE > 1 SD above age-related normal and/or eczema and/or positive skin tests); parental atopy; number of siblings; age at first wheeze; sex; serum-soluble IL-2R; proliferation of peripheral blood mononuclear cells (PBMC) to beta-lactoglobulin and to D. pteronyssinus; production of IFN-gamma on stimulation of PBMC with beta-lactoglobulin and with D. pteronyssinus. A positive clinical outcome (child requiring prophylactic antiasthma treatment after 1 yr) was observed in 53 (49.5%) children. Predictor variables were assessed by univariate and multivariate logistic regression. Wheeze was more likely to be persistent in older, atopic children with biparental atopy. The model offering best prediction of persistent wheeze with least risk of including asymptomatic subjects was age at presentation + sIL-2R. Trials of early intervention strategies using a logistic regression equation based on this model for patient recruitment can now be designed.